A REAL FIRE IN SMALL APARTMENT – A CASE STUDY

VALDIR PIGNATTA E SILVA1, RICARDO HALLAL FAKURY2, FRANCISCO CARLOS RODRIGUES2 and FABIO DOMINGOS PANNONI3

ABSTRACT Although natural fire is allowable by Brazilian standards of the fire safety design,

there are not any specifications to appliance to the ordinary buildings constructed with the typical construction and structural materials from Brazil. This paper describes the behavior of a cold formed steel framed structure in response to a real fire occurred in a residential apartment, on 3rd January, 2002. A set of residential buildings that constructed in 1995 in Limeira city, State of Sao Paulo, Brazil, is composed by four identical four-storey blocks of eight apartments per floor housed by poor people; each apartment has 44.3 m2 of total area, including two bedrooms, one living room, bathroom and kitchen. The structure and masonry were built, respectively, by cold formed steel and ordinary bricks. The Brazilian Code allows this type of buildings be neglected from structural checking in fire situations.

The fire has started in a little Christmas tree into an apartment on the fourth floor. Although the fire had been developed without any restraint, nothing spread to the neighboring apartments occurred, and not even the damaged structure presented risks of global or partial collapse. In this case study, a real fire in a residential apartment was modeled, as well the structural fire resistance; the maximum fire temperature reached during the burning on the structural material was assessed by metallographic tests and computational modelling, to the further comparative analysis between the theoretical and actual results.

1 Professor, Polytechnic School of the University of Sao Paulo, Department of Structural and Geotechnical Engineering, Av. Prof. Almeida Prado, trav2, n271, CEP 05508-900, Sao Paulo, Brazil email: valpigss@usp.br. 2 Professor, Federal University of Minas Gerais, Department of Structural Engineering, Av. Contorno, n842, 2o andar, CEP 30110-060, Belo Horizonte, MG, Brazil email: fakury@dees.ufmg.br, francisco@dees.ufmg.br 3 Specialist Engineer, Ph.D.. Gerdau Acominas S.A., Rua Cenno Sbrighi, 170, 2º andar, Ed. II, CEP 05036-010, Sao Paulo, SP, Brazil Email: fabio.pannoni@gerdau.com.br

A CFD scenario, representative from the actual building geometry, materials and fire behavior was modeled by means of the software SMARTFIRE. Details of the fire scenario as actual fire load, material composition and location of the furniture imputed to the modeling were obtained from the apartment’s owner and measurements taken during the site inspection; after that, a steel thermal analysis in fire situation was conducted by means the software Supertempcalc, and the structural analysis of the steel frame was performed according to the expressions recommended by the Brazilian Standard for design of cold formed steel for room temperature, similar to AISI/USA, but using the strength reduction factors recommended by Eurocode 3 part1.2.

. 1. INTRODUCTION

Although the Brazilian population density is not to be big (180 million inhabitants for

8,5 million m2), it experiences a massive people migration to the biggest cities, in the search for a job. Most part of these people, due to their very poor economic conditions, join or create what we call “favela” (shantytown). An alternative to reduce this situation is the construction of low-cost buildings. The cost reduction is gotten in detriment of requirements as comfort and safety, but, still, such requirements are sufficiently superior to those found in the “favela”. The Brazilian regulation allows that, buildings of social interest, with limited total area and height, can dispense structural verification in fire situation. These residential buildings are, most of all, an assemblage of small area apartments, separated by concrete or ceramic blocks. The inherent low fire load, the structure partial protection against fire and compartmentation, while imperfect, allows intuit that the safety is not very harmed in these cases.

The four storey steel framed building “Conjunto Habitacional Juscelino Kubitschek” (Fig. 1), conceived as of social interest, was erected in 1988, in Limeira city (State of Sao Paulo). It constitutes four identical blocks, two by two. Each block has four floors, with eight apartments per floor. These apartments are very simple, have 44.29 m2 total area, and contain two bedrooms, a living room, kitchen, a bath-room and a small service area. The building has a steel cold formed structure (Fig. 2).

Fig. 1 – Conjunto Habitacional Juscelino Kubitschek.

Fig. 2 – Column and beams view.

In January 2002, a fire started in the fourth floor apartment, and despite the fire not

had been effectively controlled, even so, the fire didn’t spread to other apartments. The fire destroyed all living room furniture and covering. The family who slept in the bedrooms was saved, and part of all inner masonry was demolished in the after fire due to small cracks and fissures. The fire didn’t damage the steel structure. This real fire allowed to the authors to face a more scientific evaluation of the fire effects. This paper describes the analysis

conducted and presents the main conclusions, taking as reference the real situation found at the end of the fire after.

2. THE BUILDING STRUCTURE

As seen on Fig.3, cold formed steel were used for columns and beams (i.e., box-

sections). Columns, 200 mm high x 100 mm width were constructed using two U shapes (200 x 50 x 30 x 5,0), fillet welded all around the sections. All the beams had the same external dimensions, and were constructed with two U shapes (200 x 50 x 15xt, welded the same way. The structural elements didn’t use any traditional fire protection, but they can be considered partially protected by a 100 mm masonry (concrete blocks plus a mortar-based cover), according Fig. 4. A concrete/ceramic composite slab was used, with 70 mm thickness + 50 mm top cover + 20 mm undercover, resulting in a 140 mm composite deck.

a) Cross section (b) Typical connection Fig. 3 – Beams and columns.

3. THE FIRE AND ITS CONSEQUENCES

The building plant can be seen on Fig. 4, and gives an idea of the furniture

arrangement, and the place were the fire started, as assumed by the Firemen Dept. Report(1). The composite slab started to degrade just close were the fire initiated (Fig. 5). The steel structure didn’t collapse and showed any visible damage. Masonry cover showed some local spalling, as can be seen on Fig. 5, but, in general, there were no cracks or fissures on the walls neither on the slabs, so, compartmentation was maintained. The fire didn’t spread to bedrooms or bath-room, however, the smoke production was intense; the reason was due to compartmentation, although imperfect, was efficient.

4. METALLOGRAPHIC ANALYSES

The original structural design was not found, then samples (300 mm x 60 mm) were took off from living room’s column and beams, for metallographic and tensile tests. The objectives were to obtain steel’s chemical composition and follow any tensile reduction. One sample, A1, is representative of cold regions (it was protected by walls in four sides, and the original paint was intact after wall demolishment) and served as a reference to the others (Figs. 6 and 7).

Table I gives the sample’s chemical composition, and Table II shows the measured tensile properties. It can be shown that it was used a high strength low alloy steel containing Copper and Chromium, that is, a weathering steel.

100

2003,3 to 4,75 mm – beams 5,0 mm - columns

15 mm - beams 30 mm - columns

Fig. 4 – Apartment’s plant with furniture, showing where the fire started.

Fig. 5 – Composite slab damage detail.

Fig. 6 – Samples positioning.

Fig. 7 – Central column .

The micrographs reveal a perlitic-ferritic structure, very typical for a structural grade

steel. It can be seen a very thin layer of decarburized steel, caused by the reducing atmosphere along the fire. For most hot rolled shape production, final rolling occurs when the steel is about 870°C or higher, depending the mill procedures. Austenite is the metallographic structure, that is transformed into ferrite and pearlite along cooling to ambient temperature.

The Iron-Carbon diagram can be used to foresee the steel phases changes. The diagram has three invariant points of witch only one is relevant to the present study. This is known as the eutectoid point and occurs at 0,8% carbon and 723oC. At this stage austenite will begin to transform to a constituent known as pearlite which consists of alternate plates or lamellae of ferrite and cementite.

The temperatures at which transformations take place are known as the critical temperatures. Thus, the eutectoid temperature, A1, is 723oC. If the steel temperature doesn’t exceed, for some time, this temperature, we expect that steel mechanical properties will be acceptable.

Fire started here

Sample A3

Sample A4

A1 (1,10 m do piso)

A3 (0,60 m do piso) A4 (1,30 m do piso)

Table 1 – Sample’s chemical composition. Element A1 (cold) A2 A3 A4 A6 A7 A8

%C 0,119 0,117 0,108 0,112 0,114 0,106 0,106 %Mn 0,432 0,398 0,394 0,398 0,398 0,395 0,386 %P 0,094 0,096 0,099 0,104 0,096 0,093 0,086 %S 0,012 0,008 0,013 0,014 0,008 0,007 0,006 %Si 0,453 0,437 0,480 0,493 0,435 0,426 0,415 %Al 0,045 0,060 0,056 0,057 0,060 0,059 0,058 %Nb 0,013 0,015 0,015 0,015 0,015 0,015 0,014 %V 0,006 0,005 0,006 0,006 0,005 0,005 0,005 %Ti <0,005 <0,005 <0,005 <0,005 <0,005 <0,005 <0,005 %Cu 0,204 0,200 0,219 0,215 0,199 0,200 0,201 %Cr 0,869 0,848 0,890 0,896 0,844 0,841 0,830 %Mo <0,005 <0,005 <0,005 <0,005 <0,005 <0,005 <0,005 %B 0,0001 0,0001 0,0001 0,0001 0,0001 0,0001 0,0001 %Ni 0,013 0,013 0,012 0,012 0,013 0,013 0,013

Table 2 - Tensile properties for the test specimens.

Properties

Sample Yield strength (MPa)

Ultimate Strength (MPa)

Elongation,% (200 mm)

Decarburizing thickness (µm)

Grain size

(ASTM) 1 402 529 30 50 11 2 394 517 29 200 11 3 400 525 32 120 11 4 389 507 31 120 11 6 444 557 27 40 11 7 445 558 27 Nil 11 8 406 533 28 150 11 The residual mechanical properties, after cooling to ambient temperatures, will be the

same as in the pre-fire condition. Smith et al(2) and Kirby et al(3) give experimental data for structural steels submitted to different heating – cooling cycles. Any temperature rise between 720 oC and 870 oC has a low impact on steel’s mechanical properties after cooling to ambient temperatures. Any heating beyond these values will cause a permanent transformation, resulting in a grain growth, and, sometimes, in hardening that, with the cooling, will affect adversely the residual mechanical properties. Tests carried on seven samples didn’t reveal any significant change on microstructure; the same is valid for the tensile properties. This permits us to affirm that structural components were not heated to temperatures up to 870oC. It was clear the existence of an original adherent mill scale on the central column, a strong indication that temperatures were below 550oC. Samples took from beams showed a much more rougher surface, indicating a temperature range from 650oC up to ≈ 720oC.

5. COMPUTATIONAL FLUIDS DYNAMICS BASED ON THE SOFTWARE SMARTFIRE

An educational version of the SMARTFIRE V4.0 (4) software, developed by the Fire

Safety Engineering Group at the University of Greenwich, was used to perform the fire simulations in this study, with a special feature developed particularly for this project. This

included a simultaneous capture of temperatures data from selected control volumes specially positioned along the structure.

SMARTFIRE is an open architecture CFD environment written in C++; it has four major components: a CFD numerical engine, a graphical user interface, an automated meshing tool and an intelligent control system. It permits to simulate a fire, in a fast and confident way. It uses three-dimensional unstructured meshes, enabling complex irregular geometries to be meshed without the recourse of cruder methods such as the stepped regular meshes or body-fitted meshes. The first step was to introduce the scaled apartment ambient that is materials, fire characteristics, etc. Walls, slabs, apertures, fuels, were all introduced in the model. Some control volumes were introduced, in a way of capture the gas temperatures (along time) facing the samples we took for analysis (Fig. 8). In the following, it was created a CFD mesh with more than 105,000 tetrahedron elements. We started the fire on the TV rack (as proposed by the Fire Dept. Report), propagating to a Christmas tree and, then, to a sofa. Some items were not included (as the living room curtain), for simplicity. The peak heat output inputs were 15.65 MJ for the sofa, 7.84 MJ for the Christmas tree, and 5.37 MJ for the TV rack; detais were presented before by Pannoni et al.(5). Objects burns were triggered, so some trigger volumes were constructed to activate the other fire, and so on. We created two trigger condition to ignite the fire: 573K OR maximum radiation Y negative flux greater than 22000W/m2 (the radiant flux down onto the object from the hot gases, in the ceiling layer – the most common mechanism for remote objects to be ignited). It was assumed the six-flux radiation model, and for turbulence, the K-Epsilon model. Simulation results are represented, synthetically, in Fig. 9 and 10. Maximum temperatures were obtained in the control volume located over the steel sample located at V1, that is, in the top of window. Minimum temperatures were obtained in the control volume close to sample A3, that is, in the lower part of the central column. In a general way, higher temperatures were obtained with times around 25 – 26 minutes from the fire beginning.

Fig. 8 – SMARTFIRE graphical interface, showing volume controls.

Fig. 9 – SMARTFIRE fire scenario after 2 minutes.

It is important to point out that, while the inner doors were all opened,

compartmentation was very effective. Fig. 11 shows a plane for a 5 minute scenario; temperatures around 190oC were obtained inside the bedroom (≈ 2 m height). Fig. 12 shows the gas temperatures close the central column and beam (the most heated one).

0

100

200

300

400

500

600

700

800

0 10 20 30 40 50

time (min)

tem

pera

ture

(°C

)

Fig. 10 – CFD temperature estimate, using specific control volumes.

Fig. 11 – Compartmentation effect over temperature distribution, 5 min scenario.

If it was used traditional method, ISO-Fire(6), the section factors would be 147 m-1 and

208 m-1, respectively, for column and beam, or approximately 600°C and 655°C for 15 min

and 780°C and 815°C for 30 min.

0

100

200

300

400

500

600

700

0 10 20 30 40 50

Time (min)

Tem

pera

ture

(°C)

close of the column

close of the beam

0

0.2

0.4

0.6

0.8

1

0 200 400 600 800 1000 1200

temperature (°C)

redu

ctio

n fa

ctor

s(E

C4

- cla

ss 4

)

ky θ (class 4)

kEθ

Fig. 12 - Gases temperature based on the SMARTFIRE.

Fig. 13 – Reduction factors of the steel.

6. THERMAL ANALYSES BASED ON THE SOFTWARE SUPERTEMPCALC

The reduction factors of the steel for strength and modulus of elasticity were presented in the Fig. 13. The total emissivity for steel, concrete or masonry was considered as 0.7. The density for masonry and concrete, were considered, respectively, as 1600 and 2300 kg/cm3. It was considered moisture content of 1.5 % of concrete weight; other thermal parameters from EC 2(7), EC 3(8) and OZONE (9).

Column and a 10 cm thick normal brick temperatures (facing one side) were obtained using the software Supertempcalc (10) (for convection, radiation and conduction) with ISO and Natural fires, based on item 5, Fig 12 (Figs. 13 and 15). The equivalent temperature is the uniform temperature that causes the same resistance to normal force (Figs. 14 and 16 and Table 3).

A2 A6 A7 A8 A4 A3

0 10 20 30 40 50 60 70 80 900

200

400

600

800

1000

1200

1400

Nfi,

Rd

[kN

]

Time [min]

Fig. 13 - Temperature at 30 min on the central column and wall (ISO-Fire).

Fig. 14 - Resistance to normal force vs. time on the column with wall, where Nfi,Rd = A fy ky,θ (ISO-Fire).

Table 3 - Equivalent temperature for the column – ISO-Fire Time (min) ky,θ Equivalent temperature (°C)

0 1 20 4.56 0.881642 207.86

9 0.632318 415.82 13.56 0.372497 588.98

18 0.239644 613.68 22.56 0.185231 629.94

27 0.145084 661.73 31.56 0.109844 715.99

36 0.088712 765.90 40.56 0.073862 810.99

45 0.065033 842.34

0 10 20 30 40 50 60 70 80 90700

800

900

1000

1100

1200

1300

Nfi,

Rd

[kN

]

Time [min]

Fig. 15 -Temperature at 31.5 min on the central column and wall (Natural-Fire).

Fig. 16 - Normal resistant vs. time on the column with wall (Natural-Fire).

At 31.5 min the steel temperature that causes the same resistance to a normal force is equivalent to a uniform temperature 382.5 °C (reduction factor equal 0.675).

The temperatures of the beam under the slab (Figs. 17 and 19) were determined by thermal analyses using the software Supertempcalc (convection, radiation and conduction) with ISO and Natural fires. The equivalent temperature is the uniform temperature that causes the same resistance to bending moment (Figs. 18 and 20, and Table 4).

0 10 20 30 40 50 60 70 80 900

10

20

30

40

50

60

70

MCa

p [k

Nm

]

Time [min]

Fig. 17 - Temperature at 30 min on the beam under a slab (ISO-Fire).

Fig. 18 –Resistance to bending moment vs. time on the beam under a slab (ISO-Fire).

Table 4 - Equivalent temperature for the beam – ISO-Fire

Time (min) Reduction factor Equivalent temperature (°C)

0 1 20 4.56 0.82 267.07

9 0.5 518.04 13.56 0.279 608.81

18 0.211 619.86 22.56 0.18 632.79

27 0.143 664.12 31.56 0.114 707.88

0 10 20 30 40 50 6015

20

25

30

35

40

45

50

55

60

65

MCa

p [k

Nm

]

Time [min]

Fig. 19 - Temperature at 30 min on the beam under a slab (Natural-Fire).

Fig. 20 – Resistance to bending vs. time on the beam under a slab (Natural-Fire).

At 27 min the steel temperature that causes the same bending moment resistant is equivalent to a uniform temperature 610 °C (reduction factor equal 0.27).

7. STRUCTURAL ANALYSES

The complete tri-dimensional building structure was modeled including all 4-blocks, slabs simulated as diaphragm with infinite stiffness in its plane, and with zero stiffness in the perpendicular plane. All connections between columns and beams and foundations were considered fixed in both directions. (Figs. 2 and 3). The structural analysis was done using the software SAP 2000(11), with accidental combination of loads recommended by NBR 14323 (12, 13):

∑∑==

++=m

jkfiQifiqexcQkfiGifigi

n

idfi FFFF

1,,,,,,,

1, """" γψγ (1)

Where: Ffi,d – design value of action in fire FGi,fi,k – characteristic value of permanent action i in fire FQj,fi,k – characteristic value of variable action j in fire FQ,exc. – characteristic value of thermal (exceptional, accidental) action, generally equal zero in the presence of gravitational load and ISO-Fire γgi,fi = 1,2 - partial safety factor for permanent action i in fire γq,fi = 1,0 - partial safety factor for variable action j in fire Ψ = 0.2 for places where there is neither predominance of weights of equipment that remains fixed for long periods of time, nor of people concentration. ψ = 0.0 for wind loads

The dead loads were pre-casted slab concrete/ceramic 120 mm thick and wall with normal brick 100 mm thick. The considered live loads were 1.5 kN/m2 on the floor, 0.5 kN/m2 on the ceiling and 3.0 kN/m2 on the stairs.

The cold formed steel was verified based on the NBR 14762(14), similar to AISI(15), adapted for the fire situation. This adaptation consisted in including the reduction factors ky,θ and kE,θ for class 4 elements from Eurocode 3 – Part 1.2(8). Besides that, other equations (p. ex. eqs. 2 and 3) recommended by NBR 14323(12, 13), based on the old Eurocode 3 – Part 1.2(16), were verified using a simplified software developed by Soares(17):

0,1

k W1 yxel,,

,

,,

,

, ≤

⎟⎟⎠

⎞⎜⎜⎝

⎛−

+=

yfiex

Sdfi

Sdfix

Rdfi

SdfiE

fNN

MNN

θ

ϕ and 0,1k W yxel,

,,

,

, ≤+=y

Sdfix

yy

SdfiR f

MfkA

N

θθ

ϕ (2)

Nfi,Sd - design value of the compression in fire Nfi,Rd - design value of the compression resistance of the gross cross-section, in fire Mx,fi,Sd - design value of the bending moment in fire Wel,x - plastic section modulus Nex, fi - value of elastic buckling load A – area ky,θ - yield strength reduction factor fy – characteristic value of yield strength

( )θ

ϑθ

λρ

,0

,, 1+

= yyeffRdfi

fkAN for 0 ≤ λ0,θ < 0.2 or

2.1,

,yyeff

Rdfi

fkAN ϑθρ= for λ0,θ ≥ 0.2 (3)

where: Aeff – area effective of cross section including local buckling effects ρθ - reduction factor for flexural buckling for a temperature θ λ0,θ - reduced slenderness in fire

The compartmentation efficiency and the heat transfer between steel and slab or wall lead to a thermal gradient in the structural elements. That gradient and, consequently, the efforts were not considered in this structural analysis. The actions from the axial deformation and the little minor inertia bending moment were also not considered. By other hand, we used the maximum temperatures from metallographic or thermal analysis as a uniform distribution.

In this paper, we studied the central column with one face protected by a wall and the beam close the window, the hottest region. The column buckling lengths were 2.92 m e 3.68 m. One adopted fy = 390 MPa, minimum value from metallographic.

The column temperature was considered below 550 °C, and it is based on the metallographic tests. As shown by SMARTFIRE and Supertempcalc, the medium temperature was below 400 °C. At this level of temperature, the columns of the 1st and 4th floor have ϕE and ϕR ≤ 1.0, then, there is structural safety based on the hypothesis of this paper. It is possible to notice in loco that no failure, global or local, occurred with the column.

The hottest beam temperature ranged 650°C to 723 °C, following metallographic observation. Based on the SMARTFIRE and Supertempcalc, the beam reached a medium temperature of 610°C. For that level of temperature, ϕR ≤ 1,0, i. e., the structures are in fire safety. By simplified design method no collapse would happen, confirmed by actual facts.

Using the curve ISO 834(6), and determining the equivalent temperature on the beam, based on the Supertempcalc, the beam fire resistance is approximately 30 min, i. e., ϕR ≤1,0 for FR = 30 min. 8. CONCLUSION

A small apartment real fire scenario was analyzed. A steel structure, without any fire

protection showed inherent resistance. The compartmentation, while imperfect, was made evident. Thermal and structural analysis, by using software (SMARTFIRE, Supertempcalc and others) and metallographic tests were performed. It was concluded by metal analysis that steel components were heated up to temperatures below 550°C for the central column, and around 650°C for the hottest beam. Computational modeling shows “medium” temperatures about 380°C in central column and 610°C along beam. Based on these results, and applying simplified structural analysis, it can be observed acceptable safety levels.

The sequence of this work is already in progress, that is, the study of the inherent fire safety of low-cost, residential buildings, without the traditional application of fire protection products.

9. REFERENCES

[1] Corpo de Bombeiros da Polícia Militar do Estado de São Paulo (State of Sao Paulo Firemen Dept.). Certidão de Sinistro n.015/130/03, expedido pelo 16o GB – 1o SGB – 3o/4o PB, Limeira, State of Sao Paulo. Brazil. 2003. [2] Smith, C.I., Kirby, B.R., Lapwood, D.G., Cole, K.J., and Cunningham, A.P., “The Reinstatement of Fire Damaged Steel Framed Structures”, British Steel Corporation, Teeside Laboratories. United Kingdom. 1980.

[3] Kirby, B.R., Lapwood, D.G. and Thomson, G., “The Reinstatement of Fire Damaged Steel and Iron Framed Structures”, British Steel Corporation, Swinden Laboratories. United Kingdom. 1986. [4] Ewer, J., Jia, F., Grandison, A., Galea, E. and Patel, M., “SMARTFIRE V4.0 User Guide and Technical Manual”, Fire Safety Engineering Group, University of Greenwich, UK, 2004. [5] Pannoni, F.D., Silva, V.P., Fakury, R.H., and Rodrigues, F.C., “Simulation of the Dynamics of the Fire at 41 Angelo Perillo Road, Limeira, Brazil, 2002”, Proc. XXVI Iberian Latin-American Congress on Computational Methods in Engineering – CILAMCE 2005, Guarapari, ES, Brazil, 19th – 21th October 2005. [6] International Standardization for Organization - Fire-resistance tests - Elements of building construction. ISO 834. Genève, 1994. [7] European Committee for Standadization. Eurocode 2: Design of concrete structures - Part 1-2: General rules - Structural fire design part 1.2. Brussels. 2004. [8] European Committee for Standadization. Eurocode 3: Design of steel structures - Part 1-2: General rules - Structural fire Design. Brussels. 2005. [9] Cadorin, J. F; Franssen, J. M. Software OZONE V 2.2. Liège. 2002. [10] Anderberg Y. SUPER-TEMPCALC. A commercial and user friendly computer program with automatic FEM-generation for temperature analysis of structures exposed to heat. Fire Safety Design. Lund; 1991. and TCD 5.0. User’s Manual. Fire Safety Design. Lund. 1997. [11] SAP 2000 Nonlinear. Version 8.2.7, Computers and Structures, Inc., Berkeley, CA, USA. 2003. [12] NBR 14323. Dimensionamento de Estruturas de Aço de Edifícios em Situação de Incêndio (Steel fire design) ABNT - Associação Brasileira de Normas Técnicas (Brazilian Standard). Rio de Janeiro. 1999. [13] Silva, P. V.; Fakury, R. H., “Brazilian standards for steel structures fire design”, Fire Safety Journal 37. Elsevier, 2002. [14] NBR 14762. Dimensionamento de Estruturas de Aco Constituidas por Perfis Formados a Frio (Cold formed steel design). ABNT - Associação Brasileira de Normas Tecnicas (Brazilian Standard), Rio de Janeiro, 2001. [15] American Iron and Steel Institute, “Cold Formed Steel Design”, AISI. USA, 1996. [16] European Committee for Standardization. Eurocode 3 – Part 1.2. Design of Steel Structures-Structural Fire Design, Brussels, 1995. [17] Soares, C. H., “Dimensionamento de estruturas de aço constituídas por perfis formados a frio em situação de incêndio”, (Cold formed steel design – Master Degree), UFMG, Belo Horizonte, 2002. Acknowledgements To CBCA – Brazilian Center of Steel Construction for supporting this research, to COSIPA – Companhia Siderurgica Paulista for all lab. tests, to FIPAI/EESC-USP, to Dr. John Ewer from the University of Greenwich, and Mrs.Marcia Olivieri Silverio Pannoni – MyDeWi.